Gallium nitride (GaN) based optoelectronic and electronic devices are of tremendous commercial importance. The quality and reliability of these devices, however, is compromised by high defect levels, particularly threading dislocations, grain boundaries, and strain in semiconductor layers of the devices. Threading dislocations can arise from lattice mismatch of GaN based semiconductor layers to a non-GaN substrate such as sapphire or silicon carbide. Grain boundaries can arise from the coalescence fronts of epitaxially-overgrown layers. Additional defects can arise from thermal expansion mismatch, impurities, and tilt boundaries, depending on the details of the growth of the layers.
The presence of defects has a deleterious effect on epitaxially-grown layers. Such effect includes compromising electronic device performance. To overcome these defects, techniques have been proposed that require complex, tedious fabrication processes to reduce the concentration and/or impact of the defects. While a substantial number of conventional growth methods for gallium nitride crystals have been proposed, limitations still exist. That is, conventional methods still merit improvement to be cost effective and efficient.
Progress has been made in the growth of large-area gallium nitride crystals with considerably lower defect levels than heteroepitaxial GaN layers. However, most techniques for growth of large-area GaN substrates involve GaN deposition on a non-GaN substrate such as sapphire or GaAs. This approach generally gives rise to threading dislocations at average concentrations of 105-107 cm−2 over the surface of thick boules, as well as significant bow, stress, and strain. Reduced concentrations of threading dislocations are desirable for a number of applications. Bow, stress, and strain can cause low yields when slicing the boules into wafers, make the wafers susceptible to cracking during down-stream processing, and may also negatively impact device reliability and lifetime. Another consequence of the bow, stress, and strain is that, during growth in m-plane and semipolar directions, even by near-equilibrium techniques such as ammonothermal growth, significant concentrations of stacking faults may be generated. In addition, the quality of c-plane growth may be unsatisfactory, due to formation of cracks, multiple crystallographic domains, and the like. Capability to manufacture substrates larger than 2 inches is currently very limited, as is capability to produce large-area GaN substrates with a nonpolar or semipolar crystallographic orientation. Most large area substrates are manufactured by vapor-phase methods, such as hydride vapor phase epitaxy (HVPE), which are relatively expensive. A less-expensive method is desired, while also achieving large area and low threading dislocation densities as quickly as possible.
Ammonothermal crystal growth has a number of advantages over HVPE as a means for manufacturing GaN boules. However, the performance of ammonothermal GaN crystal growth processing may be significantly dependent on the size and quality of seed crystals. Seed crystals fabricated by HVPE may suffer from many of the limitations described above, and large area ammonothermally-grown crystals are not widely available.
Lateral epitaxial overgrowth is a method that has been widely applied to improvement in the crystallographic quality of films grown by vapor-phase methods. However, no one has yet been able to apply such methods to ammonothermal GaN growth.
From the above, it is seen that techniques for improving crystal growth are highly desirable.